Supplement Article

Foods, obesity, and diabetes—are all calories created equal?

Dariush Mozaffarian

Diet has become one of the top risk factors for poor health. The incidence of cardio-
metabolic disease in the United Sates, in Mexico, and in most countries is driven
fundamentally by changes in diet quality. Weight gain has been typically framed
as a problem of excess caloric intake, but, as reviewed in this paper, subtle changes

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

in the quality of diet are associated with long-term weight gain. In order to success-
fully address obesity and diabetes, researchers and policy makers have to better un-
derstand how weight gain in the long term is modulated and to change the focus
of research and public policy from one based on counting calories to one based on
diet quality and its determinants at various levels.

INTRODUCTION these conditions should be considered a primary risk

factor for disease because each is in an intermediate
According to data from the Global Burden of Disease marker, a downstream consequence of poor lifestyle.3
2010 Study, which looked at the modifiable causes of Thus, for cardiometabolic disease, the fundamental risk
death in 187 countries, dietary risks are now the leading factors at the individual level are poor diet, physical in-
cause of poor health in the United States, in Mexico, activity, and smoking (Figure 2).
and in almost every country around the world.1 Data In the last 20–30 years, lifestyles, including dietary
from the United States shows that most of the deaths at- habits, have changed across the world, accompanied by
tributable to cardiovascular and circulatory diseases, as a global obesity epidemic. While physical activity has
well as some cancers, are related to diet (Figure 1).2 decreased in many regions, especially in low-income
With few exceptions, in most countries, dietary risk countries, in high-income countries such as the United
now exceeds tobacco smoking as the leading cause of States the best available evidence suggests that overall
poor health and, therefore, represents one of the most physical activity has remained stable or even increased
pressing public health challenges of the 21st century. over the last 30 years as the obesity epidemic has
mounted. This indicates that the main driver of the obe-
Fundamental cardiometabolic risk factors sity epidemic in the United States is worsening diet,
while in most low-income countries it is likely a combi-
When considering cardiometabolic risk, it is crucial to nation of decreased physical activity and worsening
consider the continuum of risk factors Over time, death diet.1,2
is the worst outcome, preceded by a range of diet- Because diet is a fundamental risk factor for obe-
related clinical outcomes such as myocardial infarction, sity, it is crucial to understand the pathways through
stroke, heart failure, cognitive decline, renal failure, and which it increases weight. The current conventional
peripheral arterial disease, among others. Intermediate wisdom is that obesity reflects a problem of calories and
risk factors for these conditions include high choles- energy balance: people are eating too many calories and
terol, high blood pressure, and diabetes, and the under- exercising too little. This view has driven both scientific
lying drivers of these include obesity, endothelial investigations and public policies. For instance, for de-
dysfunction, and metabolic dysfunction. Yet, none of cades, low-fat diets were recommended for obesity
Affiliation: D. Mozaffarian is with the Friedman School of Nutrition Science & Policy, Tufts University, Boston, Massachusetts, USA.
Correspondence: D. Mozaffarian, Friedman School of Nutrition Science & Policy, Tufts University, 150 Harrison Ave, Boston, Massachusetts
02111, USA. E-mail: [email protected]. Phone: þ1-617-636-0374.
Key words: cardiometabolic risk, diet quality, obesity, weight change.
C The Author(s) 2016. Published by Oxford University Press on behalf of the International Life Sciences Institute.
V
All rights reserved. For Permissions, please e-mail: [email protected].

doi: 10.1093/nutrit/nuw024
Nutrition ReviewsV Vol. 75(S1):19–31
R
19
 Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024
Figure 1 Leading risk factors for disease-related death and disability-adjusted life yeares in the United States, 2010. Reproduced
from US Burden of Disease Collaborators (2013)2 with permission.

prevention, simply due to the higher energy content per acts on obesity predominantly through worsening diet,
gram of fat (9 kcal/g) compared with carbohydrates and rather than decreased physical activity, by changing the
proteins (4 kcal/g). quality and quantity of food eaten, increasing the fre-
Certainly caloric intake has risen in most of the quency of snacking in front of the TV, and influencing
world, but why has this been so? Has human biology food choices through marketing. Another relevant fac-
changed to make us hungrier or more weak-willed? Not tor appears to be the quality and quantity of sleep. Over
likely. Randomized trials and prospective cohort studies time in recent decades, the average hours of sleep per
have advanced nutrition science tremendously, building night have decreased, and growing evidence supports a
a new view of diet, obesity, and diabetes that must enter role of insufficient sleep on both the quantity and qual-
clinical practice and public policy.4 This modern evi- ity of food consumed.11–13 Together, these findings
dence suggests that a major driver of increased caloric each support a reduction in diet quality as one root
intake across diverse populations has been a decrease in cause of the modern obesity epidemic.14
the quality of diet: changes in the types and quality of Notably, overall food quality and dietary patterns
foods consumed, which together influence long-term also have a major impact on metabolic risk independent
energy balance.5–8 Some of the key factors that charac- of obesity. In other words, a person can remain obese
terize diet quality include carbohydrate quality; intakes and substantially decrease their metabolic risk with a
of whole foods such as nuts, beans, fruits, and vegeta- healthy diet, while a lean person can develop substantial
bles; specific fats and oils; and overall dietary patterns. metabolic dysfunction with a poor diet. In addition, in-
For example, in Mexico, so-called traditional dietary dependent of obesity status and total calories, different
patterns, characterized by the consumption of corn tor- dietary factors influence visceral adiposity and liver fat
tillas and maize-based preparations, beans, vegetables, production (hepatic de novo lipogenesis). In particular,
fruits, and vegetable-based fats, have been associated rapidly digestible carbohydrates activate de novo lipo-
with lower risk of obesity and diabetes.9,10 genesis and influence adiposity, as do trans fats. Recent
A second likely contributing factors is increased controlled trials further suggest, at least to 1 month,
television watching. Importantly, randomized trials in that the quality of the diet may actually influence meta-
youth demonstrate that increased television watching bolic expenditure.15 While longer-term trials are needed

20 Nutrition ReviewsV Vol. 75(S1):19–31

R
 Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024
Figure 2 Risk factors for cardiometabolic disease. Reproduced from Mozaffarian et al. (2008)3 with permission.

to confirm this last finding, these new results represent nonobese adult gains less than half a kilogram per
a potential paradigm shift – quality of the diet influ- year—a small, subtle amount of weight gain.16 Yet, this
ences not only energy intake, but also energy output. small weight gain adds up over time, contributing to
the obesity epidemic now occurring across the world.
Diet quality and weight gain. Several experimental stud- This gradual weight gain highlights 2 crucial con-
ies have looked at short-term weight loss in obese peo- siderations. First, because the change is so subtle, it is
ple. In the short-term, over several months, a person difficult for most people to perceive what is causing the
can lose weight on almost any diet, simply by paying at- weight gain. Second, this remarkably small weight gain,
tention to how much they eat and dramatically decreas- in the midst of generally toxic food environments, dem-
ing their caloric intake. Thus, for short-term, rapid onstrates that our body’s intrinsic regulatory mecha-
weight loss, calories are a major player—this is why nisms are actually remarkably good at maintaining our
nearly any “fad” diet will work, at least for a while. weight. These findings suggest that environmental
Yet, if the obesity epidemic is to be meaningfully changes over the past 30 years are causing subtle
reduced, clinical care and policy priorities must shift changes in these homeostatic mechanisms, which over
away from rapid weight reduction toward prevention of the long-term are ultimately driving the obesity
long-term weight gain. Just as there was a shift in epidemic.
cardiology from focusing on better survival following In order to investigate long-term changes in weight
acute myocardial infarction (secondary prevention) to in people who were not obese at baseline, a longitudinal
preventing heart attacks in the first place (primary investigation was conducted that included over 120 000
prevention), a similar shift in focus is needed for men and women from 3 different cohort studies.16
obesity. In the case of obesity, this means moving away Individuals were followed for 20 years with repeated
from thinking mainly about treating obese patients measures of diet and weight. The analysis focused on
with weight loss (secondary prevention) and toward assessing weight change and long-term weight gain in
preventing long-term weight gain in everyone (primary relation to increasing intake of different foods. If all cal-
prevention). ories similarly caused long-term weight gain, then in-
Prevention of long-term weight gain occurs very creased intake of any food should be associated with
slowly and over many years, making it challenging to weight gain, while decreased intake of any food should
study. For instance, in the United States, the average be associated with weight loss. Yet, the findings were

Nutrition ReviewsV Vol. 75(S1):19–31

R
21
 Weight change each four years (lbs) Weight change associated with
each increased daily serving of:
-1 -0.5 0 0.5 1 1.5 2 2.5
Foods
Potato chips
Potatoes/fries
Processed meats
Unprocessed red meats
Butter
Sweets and desserts

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

Refined grains
Cheese
NHS (women)
Vegetables
NHS2 (women) Nuts

HPFS (men) Whole grains

Fruits
Yogurt

Beverages
Sugar-sweetened beverages
Alcohol
100% fruit juice
Among 120 877 men and women
followed for 20 years, adjusted every Low fat or skim milk
4 years for age, baseline BMI, sleep, Whole fat milk
and changes in activity, smoking, TV
watching, and all dietary factors Diet (zero calorie) soda
simultaneously.

Figure 3 Weight change over time associated with increased daily serving of different foods. Abbreviation: BMI, body mass index. Data
from Mozaffarian et al. (2011).16

very different. After adjustment for multiple other risk fac- For any individual food or beverage, the observed
tors and all dietary change simultaneously, as dietary in- effects were small. For example, for every 1 serving per
takes of several foods increased, weight gain was seen as day increase in sweets or refined grains, individuals
expected. In contrast, increased intakes of some other foods gained, on average, about 0.3 kg more every 4 years.
were not associated with weight gain or weight loss. And, However, while effects of individual foods were subtle,
most intriguingly, increased intakes of some foods were ac- across multiple foods they added up, largely explaining
tually associated with relative weight loss (Figure 3). the observed population weight gain over time.
The foods most associated with weight gain were Foods that were relatively neutral in terms of long-
potato chips, potatoes (including boiled, baked, and term weight gain included cheese, both low-fat and
mashed potatoes), meats, refined grains, and sweets and whole-fat milk, and diet soda. Increased intake of these
desserts. An important finding was that the weight gain foods did not result in weight gained or lost. And, fi-
associated with sweets and desserts was about the same nally, there were several foods for which increased in-
as weight gain associated with refined grains. In other take was associated with less weight gain. These foods
words, similar weight gain was seen whether one ate a included vegetables, nuts, whole grains, fruits, and yo-
serving of candy or of white bread. This is consistent gurt. These foods are unlikely to violate the laws of ther-
with metabolic evidence showing that the rapidity of modynamics. Instead, these findings provide evidence
digestion and metabolic responses of white bread are that different foods have very different effects on our
similar to that of sugar.17 Among beverages, sugar- body’s usual, powerful regulatory and compensatory
sweetened drinks, alcohol, and 100% fruit juice are each pathways. For instance, foods such as potatoes and soda
positively associated with weight gain. From this list, appear to harm these regulatory pathways, so that their
one can see that most of the foods associated with calories are not fully compensated for in long-term eat-
weight gain are rich in refined carbohydrates. ing habits, causing weight gain over time. For other

22 Nutrition ReviewsV Vol. 75(S1):19–31

R
 8
*
6
Weight Due to abdominal
gain (visceral) fat
4
(% change)
2

0
cis-18:1 trans-18:1

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

Randomized controlled trial among green monkeys fed calorie-
controlled diets containing cis or trans fat (8% energy) for 6 years.

Figure 4 Effect of fat quality on long-term weight gain in monkeys. Data from Kavanagh et al. (2007).18

foods such as cheese or milk, our bodies appear to ap- carbohydrate diet experienced a nearly 60% reduction
propriately compensate for any increase or decrease in in hepatic fat, as measured by nuclear magnetic reso-
their intake by similarly decreasing or increasing calo- nance imaging, compared with only a 25% reduction in
ries from other foods. And for other foods such as vege- hepatic fat for individuals on the low-fat diet (Figure 5).
tables, nuts, whole grains, fruits, and yogurt, our bodies This study shows that, calorie for calorie, weight for
appear to overcompensate when more calories are con- weight, there is a differential effect of different foods on
sumed, resulting in decreased energy intake over the metabolism and body function.
course of a year. Lennerz et al.21 carried out a trial to test uncon-
scious brain reward activation by comparing 2 similar
Beyond fullness and digestion: effects of different liquid shake meals. Previous trials studying brain re-
foods on weight and body function ward activation compared very different types of foods,
e.g., chocolate vs broccoli, and found that, compared
Recently, several experimental studies have begun to with less palatable foods, hedonistic foods such as choc-
elucidate the mechanisms by which different foods af- olate light up our brain reward centers, leading to the
fect weight gain. While hunger and satiety likely play a concept of food craving or even food addiction. In con-
role, there is more than just the sensation of fullness or trast, Lennerz’s experiment used 2 meals with identical
digestion time at play. Recent years have seen an explo- appearance, taste, macronutrients, calories, palatability,
sion of research in this area. Many trials show that, cal- and sweetness. The sole difference was their glycemic
orie for calorie, different foods have different effects, index: the rapidity of digestion and blood glucose re-
not only on hunger and fullness but also on insulin, sponse. This trial demonstrated significantly greater
adrenaline, other hormonal responses, de novo lipogen- postprandial activation of the brain regions involved in
esis in the liver, activation of brain reward, microbiome reward and cravings after the higher glycemic index
(gut microbiota) interactions, and even, as already men- meal, compared with the lower glycemic index meal.
tioned, possibly metabolic expenditure. Again, calorie for calorie, a change in glycemic index
For instance, Kavanagh et al.,18 found that in non- alone made people hungrier and differentially activated
human primates fed energy-matched diets with 8% en- their unconscious brain reward centers (Figure 6).21
ergy from either cis or trans fats for 6 years, the Diet quality also alters microbiome function. In the
monkeys in the trans fats group gained more weight, observational investigation of long-term weight gain de-
largely due to increased abdominal or visceral fat scribed above, yogurt was associated with less weight
(Figure 4). Subsequently, other experiments have shown gain.16 In contrast, cheese and milk were each neutral.
that trans fats activate de novo lipogenesis in the liver.19 Because a major difference between these dairy products
In other words, calorie for calorie, trans fats appear to is the active bacterial cultures in yogurt, it was hypothe-
increase hepatic fat synthesis and visceral adiposity. sized that an interaction with the microbiome might ex-
Browning et al.20 randomized 18 individuals with plain the lower weight gain. Since that study, several
nonalcoholic fatty liver disease to two groups, each with animal experiments have shown that yogurt and active
a similar goal of 5% weight loss over 2 weeks, but probiotics in yogurt interact with the microbiome to re-
achieved with either a low-carbohydrate or a low-fat duce weight gain—without any changes in total calories
diet. Even with similar calories consumed and 5% consumed.22 In one study, this interaction appeared to
weight loss, as intended, individuals on the low- be dependent on gut CD4 cells and interleukin-10

Nutrition ReviewsV Vol. 75(S1):19–31

R
23
 Low-calorie Low-carbohydrate
diet (n = 9) diet (n = 9)
• 18 subjects with NAFLD
• Controlled, equal 5% weight loss with
low-calorie vs low-carb diets over
2 wk
• Hepac fat measured by H-NMR

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

Similar 5% weight loss
Nearly 2-fold
greater reducon
in hepac steatosis

*
P = 0.008

Figure 5 Effect of diet quality on hepatic fat synthesis in humans. Abbreviations: NAFLD, nonalcoholic fatty liver disease; TG, triglyceride.
Data from Browning et al. (2011).20

(Figure 7).22 So, diet quality appears to have very complex effect of diet quality on energy expenditure, at least over
effects on physiology and metabolism. 1 month. Currently, the researchers are following this
The quality of the diet may also alter energy expen- up with longer-term studies to see if these results are
diture. Ebbeling et al.15 conducted a prospectively de- maintained. Nonetheless, this is remarkable: a daily
signed trial, based on a prior post hoc analysis of trials, 300-calorie difference in metabolic expenditure in an
to test the idea that modifying the quality of diet would obese person trying to maintain weight loss, based just
alter metabolic expenditure. After study participants on changing diet quality alone, could have a great im-
achieved 10% weight loss, they were randomized, in pact on any attempt to control weight.
crossover fashion, to 3 different diets varying in carbo-
hydrate quality and quantity for 4 weeks each. Diet quality, foods, and risk of diabetes. Independent of
Metabolic expenditure was assessed using doubly la- weight, diet quality has large effects on metabolic risk.
beled water, and physical activity using multiple accel- For instance, sugar-sweetened beverages both increase
erometers. After weight loss, consuming a low-fat diet obesity and, separately, increase risk of diabetes
during the maintenance phase reduced total daily en- (Figure 8).23 Recently, a global modeling study esti-
ergy expenditure by about 400 calories. This finding mated the numbers of annual excess deaths related to
had been shown many times before, leading to the con- sugar-sweetened beverage intake on global, regional,
cept of the “weight thermostat” or weight “set point”: and national scales. In Mexico, for example, about
i.e., after weight loss, the body’s metabolic rate de- 24 000 excess deaths per year were estimated to be re-
creases, causing people to get tired and cold, with high lated to sugar-sweetened beverage intake, caused by dia-
risk of weight regain. Yet, when the same individuals betes, heart disease, and obesity-related cancers.24
were fed a low-glycemic-load diet (with slowly digested When controlling for energy intake, other foods in
carbohydrates), their metabolic expenditure only de- the diet also appear to influence the risk of diabetes. For
creased by about 250 calories per day. When they were example, Pan et al.25 found that for every 100 grams of
fed a high-fat, Atkins style diet, their metabolic expen- processed meats consumed per day, a more than 2-fold
diture decreased by only about 100 calories per day. In higher risk of incident diabetes was observed (relative
other words, the same individuals, eating calorie- risk [RR], 2.28; 95% confidence interval [CI], 1.56–3.35)
controlled diets for just 4 weeks each, experienced sig- (Figure 9).
nificant differences in metabolic expenditure of nearly Similarly, Afshin et al.26 found that, adjusting for
300 calories per day. This was not due to differences in calories, for every 4 servings of nuts per week, a 13%
physical activity, which was unchanged based on accel- lower risk of incident diabetes is seen (RR, 0.87; 95%
erometers. This landmark study demonstrates a direct CI, 0.81–0.94) (Figure 10).

24 Nutrition ReviewsV Vol. 75(S1):19–31

R
 Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024
Figure 6 Effect of dietary glycemic index on brain reward activation in men. Abbreviation: GI, glycemic index. Reproduced from Lennerz
et al. (2013)21 with permission.

Sluijs et al.27 performed a prospective analysis of Focus on diet quality put into practice: the example of
340 234 participants of the European Prospective carbohydrate quality
Investigation into Cancer and Nutrition cohort and iden-
tified over 12 000 incident cases of type 2 diabetes. In this Based on the findings described above, carbohydrate
cohort, consumption of milk showed no significant associ- quality is a major contributor to long-term weight gain
ation with diabetes – a neutral effect consistent with that and risk of diabetes. Therefore, an important issue for
previously observed for weight gain in other cohorts. In health promotion is to clearly define carbohydrate quality.
contrast, however, these researchers observed a trend to- However, in order to assess carbohydrate quality, several
ward lower risk of diabetes with yogurt consumption and independent aspects must be taken into consideration: the
a significantly lower risk with cheese consumption (Figure fiber content, the whole grain content, the glycemic re-
11). Because a similar lower risk was seen for fermented sponse, and the form (liquid or solid). Each of these can
milk, but not regular milk, the findings for cheese could be altered independently, and each appears to influence
relate to its fermentation, which suggests a possible influ- health. Ideally, we should all eat mostly true whole grains,
ence of fermentation-produced vitamin K2 or intervention which induce low glycemic responses, are high in fiber,
with the gut microbiome. Clearly, further study of poten- and in solid form. Unfortunately, few common foods
tially relevant pathways is needed. Of note, while this anal- meet these criteria (Figure 12).28
ysis was observational, its findings would unlikely be To understand the best, simple “rule of thumb” for
attributed to confounding, as people who eat more cheese choosing higher-quality carbohydrates, one analysis
tend to have worse, not better, average diets and lifestyle. compared 5 different recommended metrics. These in-
Many short-term metabolic trials have confirmed cluded an industry-sponsored “whole grain stamp,” 3
the independent effects of diet quality. When people eat different US Department of Agriculture–recommended
healthier dietary patterns that contain fewer processed definitions (all based on evaluating the ingredients list),
meats, fewer refined carbs, and more fruits, vegetables, and, lastly, the ratio of total carbohydrates to fiber per
nuts, and dairy products, multiple metabolic risk factors serving, which attempts to capture the balance of overall
are substantially and rapidly improved, including starch plus sugar vs whole grain.29 This investigation
blood pressure, blood lipids, and glucose and insulin found that the best overall metric for choosing a higher-
resistance. quality carbohydrate (based on calories, fiber, sugar,

Nutrition ReviewsV Vol. 75(S1):19–31

R
25
 A
Less weight gain with either probioc
yogurt or purified Lactobacillus reuteri.

No changes in microbial composion or

total calorie consumpon.
B
Eang lactobacillus reuteri triggered
changes in microbiome funcon and

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

host immune system.

Dependent on both CD4+ T cells and

interleukin-10.
C D

Figure 7 Effect of probiotics on microbiome function in mice. Reproduced from Poutahidis et al. (2013)22 with permission.

Percent
increase
in risk of
diabetes

Risk for 1+ beverage (12 oz)/d, compared with <1/mo

Among 91 249 women followed for 8 years.

Figure 8 Effect of sugar-sweetened beverages on risk of diabetes. Abbreviation: BMI, body mass index. Data from Schulze (2004).23

sodium, and trans fat contents) was the ratio of total cereals) over a much healthier, high fiber, whole grain
carbohydrate to fiber. According to this metric, a good- cereal with a bit of added honey.
quality carb should contain, for every 10 g of total car-
bohydrate, at least 1 g of fiber (i.e., a ratio of 10:1 or Setting effective dietary priorities for policy and
lower). As a simple rule, this ratio works better than public health
other metrics such as added sugars, which could mis-
leadingly result in choosing highly refined, starchy Based on the modern nutritional evidence, in order to
products with no added sugar (e.g., many breakfast prevent chronic diseases including cardiovascular

26 Nutrition ReviewsV Vol. 75(S1):19–31

R
 Study ID Relative risk (95% CI)

Song (2004)27

Montonen (2005)28

Schulze (2007)30

Steinbrecher (2010)M, 31
Processed
Steinbrecher (2010)F, 31
meats
Mannisto (2010)32 (per 100 g/d)
HPFS

NHS I
RR = 2.28
(1.56, 3.35)

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

NHS II

Overall (I-squared = 94.3%, P = 0.000)

Note: Weights are from random effects analysis

Figure 9 Processed meat and incident diabetes. Abbreviations: CI, confidence interval; HPFS, Health Professionals Follow-Up Study; NHS I,
Nurses’ Health Study; NHS II, Nurses’ Health Study II; RR, relative risk. Reproduced from Pan et al. (2011)25 with permission.

Study ID RR (95% CI)

48
Villegas (2008) 64 191

47
Pan (2013) 79 893

46
Montonen (2005) 4304

37
Salas-Salvado (2014) 3541

45
Kochar (2010) 20 224

47
Pan (2013) 58 063

Overall (I-squared = 21.8%, P = 0.269)

Note: Weights are from random effects analysis

RR for diabetes per 4 servings (28.4g)/wk of nuts

Figure 10 Nuts and incident diabetes. Abbreviations: CI, confidence interval; FMCHES, Finnish Mobile Clinic Health Examination Survey; NHS I,
Nurses’ Health Study; NHS II, Nurses’ Health Study II; PHS, Physician’s Health Study; PREDIMED, Prevencion con Dieta Mediterranea; RR, relative
risk; SWHS, Shanghai Women’s Health Study. Reproduced from Afshin et al. (2014)26 with permission.

diseases, diabetes, and even obesity, the prevailing focus paradoxical choices and policies. For example, a sand-
has to shift away from calorie counting and toward wich with low-fat deli meats is often considered a
overall diet quality and food patterns. Rather than just healthy choice, yet it typically consists of processed
telling people to decrease calories, public health efforts bread, processed meat, and processed cheese, each high
should be aimed at increasing the proportion of calories in salt and potentially other harmful additives.
consumed from healthy foods, such as fruits, vegetables, Similarly, many low-fat products are marketed as
beans, nuts, seeds, fish, vegetable oils, plain yogurt, and healthy but contain abundant refined starch and sugar.
minimally processed whole grains. At the same time, Baked potato chips, chocolate fat-free milk, turkey sau-
actions must be taken towards limiting calories from sages, and fat-free salad dressing are all poor choices for
harmful foods, including processed meats, sweetened health. Yet, the US school lunch program recently
beverages, excess alcohol, and foods rich in starches, re- banned whole milk, but it allows sugar-sweetened choc-
fined grains, sugars, trans-fats, and salt.14 olate skim milk, at least partly due to the rationale that
The historical emphasis on total calories, total fat, chocolate skim milk has fewer total calories. Another
and individual nutrients can lead to misleading and example is mandated labeling of total calorie contents

Nutrition ReviewsV Vol. 75(S1):19–31

R
27
 Ptrend = 0.50

Ptrend = 0.06 Ptrend = 0.01

RR of
diabetes

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

Quintiles of consumption

340 234 Europeans, 8 countries, 12 403 cases of incident diabetes

Figure 11 Dairy foods and incident diabetes. Abbreviation: RR, relative risk. Data from Sluijs et al. (2012).27

Figure 12 Evaluation of carbohydrate quality. Abbreviation: LDL, low-density lipoprotein. Reproduced from Mozaffarian (2005)28 with
permission.

on restaurant menus. The implicit message of such a manufactures, may like to promote this notion. Many
policy is that a person can examine a range of menu factors can influence choice at the individual level, in-
choices and judge their healthfulness based on their cal- cluding education, income level, health status, nutri-
orie contents. Clearly, this is not the case: consuming tional knowledge, cooking skills, and personal habits
300 calories from nuts is a much healthier choice, in- such as television watching and sleep. Furthermore,
cluding for obesity, than 200 calories of soda or refined these individual drivers are then influenced by many
starch. Yet, with a total-calorie focus, consumers are be- other factors, such as social and cultural norms, social
ing driven to these choices. Modern nutrition science support, social class, and race/ethnicity. The surround-
indicates that foods should not be judged by calorie or ing physical environment can further alter eating pat-
fat content alone. terns: e.g., the foods available at workplaces, schools,
supermarkets, and restaurants; proximity and accessi-
Root causes of poor diet quality bility to these outlets; access to transportation; and
neighborhood socioeconomic status. At the macro level,
Ultimately, to address diet quality, the roots of poor agriculture, market forces, land use, transportation,
diet quality must be identified and addressed. Diet se- food production and distribution, issues of food safety,
lections are not merely based on individual choice, al- industry incentives, lobbying, marketing, and the media
though many, including major food and beverage can all impact food choices. Governments are also

28 Nutrition ReviewsV Vol. 75(S1):19–31

R
 Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024
Figure 13 Barriers and opportunities for healthy eating. Reproduced from Afshin et al. (2014)30 with permission.

also opportunities for intervention to improve the food

supply and food system.
To move forward, lessons from past public health
successes are useful. The fight against tobacco is often
described as an analogy. Yet, while similarities exist,
particularly in the many similar tactics of large food
companies and tobacco companies, this is perhaps not
the best comparison. With tobacco, there is no need for
the product: it is a fight to the death. In contrast, the
food industry, including agricultural producers, manu-
facturers, and retailers, is needed in order to feed the
world. Government, scientists, and society must help,
and impel whenever necessary, the food industry to
make food that is both healthier and profitable.
Figure 14 Lessons from past public health success. Abbreviation:
A more relevant public health analogy may be
VMT, vehicle miles traveled. Reproduced from US Centers of
Disease Control and Prevention (1999)31 with permission.
found in the automotive industry. In the United States,
between 1925 and 1995, the number of deaths per vehi-
cle mile traveled were reduced by 90%, representing
important actors; diet is shaped by their priorities, agri- one of the greatest public health successes of the 20th
cultural policies, food assistance programs, school lunch century (Figure 14).31 How was this achieved?
policies, and the healthcare system. Finally, global influ- Obviously not by focusing only on individual choices:
ences are relevant, such as multinational corporate lob- telling drivers to drive safely, or placing warning labels
bying, climate change, scientific research, and trade on cars about the dangers of driving. If policies had fo-
agreements (Figure 13).30 Based on these myriad influ- cused only on individual drivers, the gains in public
ences, the notion that diets are an individual choice is a health would have been small. Yet, this is how many nu-
fallacy. These influences each represent challenges, yet trition policies are presently focused—e.g., education,

Nutrition ReviewsV Vol. 75(S1):19–31

R
29
 Media and • Sustained, focused media campaigns, especially combined with
educaon mulcomponent strategies, focused on specific foods or drinks.
Labeling and • Mandated nutrion facts, front-of-pack labels/icons, or menu labeling to
informaon influence industry behavior and product formulaons.
• Mulcomponent diet and acvity program including classes, teacher
Schools
training, supporve policies, environmental changes, family components.
• Comprehensive worksite wellness programs for diet, acvity, tobacco.
Workplaces • Increased availability of healthier opons and/or strong nutrion standards,
combined with on-site prompts, labels, or icons.

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

• Subsidy strategies to lower prices of more healthful foods and beverages.
Economic • Tax strategies to increase prices of less healthful foods and beverages.
incenves • Long-term changes in agricultural and related policies for infrastructure to
facilitate producon, transportaon, and markeng of healthier foods.

• Restricons on ads/markeng of less healthy foods/drinks to children on

Bans and television, and near schools and public places, and on packages.
mandates
• Direct bans (e.g., sodium, trans fat) or mandates (e.g., vegetable oils).

Figure 15 Evidence-based policy interventions for diet. Data from Mozaffarian et al. (2012).33

labeling, guidelines—as though the problem is largely optimism that it is possible to meaningfully reduce the
individual consumers making unhealthy choices. burdens of disease from poor diet in a much shorter
The tremendous public health success of reducing timeframe. What is required is a focus on evidence-
driving-related deaths was achieved not by focusing on based dietary priorities, in particular overall foods and
individual choice, but by a multicomponent strategy diet quality, and on evidence-based policy priorities.
that addressed the driver, the car, the road, and the cul- Many evidence-based policies are now available to be
ture. At the individual level, this included driver educa- implemented and improve the quality of diet.33 These
tion programs, licensing, and laws aimed at regulating include sustained media campaigns, multicomponent
behavior such as using seat belts and child seats, limit- comprehensive school and workplace programs, robust
ing phone use, and (for youth) limiting driving late at economic incentives, quality standards for additives
night or with friends. Even more was done to address and for marketing to children, and long-term agricul-
the car (the product): active protections such as seat ture policies (Figure 15).
belts, child seats, motorcycle helmets, etc.; passive pro- Shifting the focus from calories and individual nutri-
tections such as padded interiors, collapsible steering ents to foods and diet quality and redirecting efforts to-
columns, shatterproof windshields, airbags; careful ward population interventions to address the root causes
crash safety standards; safety inspections; and more. of poor diet and lifestyle can meaningfully reduce the
Numerous initiatives also addressed the road (the envi- global pandemics of diet-related diseases and disability.
ronment), including road engineering, guard rails,
speed limits, stop signs, stoplights, caution signs, and so
on. And concerted actions were taken to change the Acknowledgments
culture, especially around drunk driving, such as desig-
nated driver campaigns, drunk driving legislation, and The articles in this supplement were presented as part
private advocacy about drinking and driving. of the Tenth Nestlé Nutrition Conference on Research
Together, these successful efforts provide a road- Perspectives for Prevention of Diabetes: Environment,
map for improving the food system: multicomponent Lifestyles and Nutrition, held in México City on
policies and efforts must address the consumer, the November 12 and 13, 2014. The conference was orga-
product, the food environment, and the culture of un- nized by the Nestlé Nutrition Fund of the Mexican
healthy eating.32 In the United States, about 70 years Health Foundation and the National Institute of
were needed to substantially reduce the number of car Medicine and Nutrition Salvador Zubiran. The supple-
accident deaths and it took 50 years to substantially re- ment coordinators are Ernestina Polo-Oteyza, Mexican
duce smoking. These lessons provide reasons for Health Foundation and Héctor Bourges-Rodrıguez and

30 Nutrition ReviewsV Vol. 75(S1):19–31

R
Carlos Aguilar-Salinas, National Institute of Medicine 11. Nedeltcheva AV, Kilkus JM, Imperial J, et al. Sleep curtailment is accompanied by
increased intake of calories from snacks. Am J Clin Med. 2009;89:126–133.
and Nutrition Salvador Zubiran, México. 12. Hart CN, Carskadon MA, Considine RV, et al. Changes in children’s sleep duration
on food intake, weight, and leptin. Pediatrics. 2013;doi:10.1542/peds.2013–1274.
13. Dashti HS, Scheer FA, Jacques PF, et al. Short sleep duration and dietary intake:
Funding. The conference and this supplement were epidemiologic evidence, mechanisms, and health implications. Adv Nutr.
2015;6:648–659.
funded by the Nestlé Nutrition Fund of the Mexican 14. Mozaffarian D, Appel L, Van Horn L. Components of a cardioprotective diet new
Foundation for Health. The author received a speaker’s insights. Circulation. 2011;123:2870–2891.
15. Ebbeling C, Swain J, Feldman H, et al. Effects of dietary composition on energy ex-
honorarium and travel reimbursement from the penditure during weight-loss maintenance. JAMA. 2012;307:2627–2634.
Mexican Foundation for Health for participating in this 16. Mozaffarian D, Hao T, Rimm E, et al. Changes in diet and lifestyle and long-term
weight gain in women and men. New Engl J Med. 2011;364:2392–2404.
conference. 17. Ludwig DS. The glycemic index: physiological mechanisms relating to obesity, dia-
betes, and cardiovascular disease. JAMA 2002;287:2414–2223.
18. Kavanagh K, Jones K, Sawyer J, et al. Trans fat diet induces abdominal obesity and

Downloaded from https://academic.oup.com/nutritionreviews/article/75/suppl_1/19/2797600 by guest on 16 February 2024

changes in insulin sensitivity in monkeys. Obesity. 2007;15:1675–1684.
Declaration of interest. The author reports ad hoc hono- 19. Hebbard L, George J. Animal models of nonalcoholic fatty liver disease. Nat Rev
raria or consulting from Boston Heart Diagnostics, Gastroenterol Hepatol. 2011;8:35–44.
20. Browning J, Baker J, Rogers T, et al. Short-term weight loss and hepatic triglycer-
Haas Avocado Board, Astra Zeneca, GOED, DSM, and ide reduction: evidence of a metabolic advantage with dietary carbohydrate re-
Life Sciences Research Organization; and chapter royal- striction. Am J Clin Nutr. 2011;93:1048–1052.
21. Lennerz B, Alsop D, Holsen L, et al. Effects of dietary glycemic index on brain re-
ties from UpToDate. gions related to reward and craving in men. Am J Clin Nutr. 2013;98:641–647.
22. Poutahidis T, Kleinewietfeld M, Smillie C, et al. Microbial reprogramming inhibits
Western diet-associated obesity. PLoS One. 2013;8:68596.
23. Schulze M, Manson J, Ludwig D, et al. Sugar-sweetened beverages, weight gain,
and incidence of type 2 diabetes in young and middle-aged women. JAMA.
REFERENCES 2004;292:927–934.
24. Singh G, Micha R, Khatibzadeh S, et al. Estimated global, regional, and national
1. Lim S, Vos T, Flaxman A, et al. A comparative risk assessment of burden of disease disease burdens related to sugar-sweetened beverage consumption in 2010.
and injury attributable to 67 risk factors and risk factor clusters in 21 regions, Circulation. 2015;132:639–666.
1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. 25. Pan A, Sun Q, Bernstein A, et al. Red meat consumption and risk of type 2 diabe-
Lancet. 2012;380:2224–2260. tes: 3 cohorts of US adults and an updated meta-analysis. Am J Clin Nutr.
2. US Burden of Disease Collaborators. The state of US health, 1990–2010: burden of 2011;94:1088–1096.
diseases, injuries, and risk factors. JAMA. 2013;310:591–606. 26. Afshin A, Micha R, Khatibzadeh S, et al. Consumption of nuts and legumes and
3. Mozaffarian D, Wilson P, Kannel W. Beyond established and novel risk factors life- risk of incident ischemic heart disease, stroke, and diabetes: a systematic review
style risk factors for cardiovascular disease. Circulation. 2008;117:3031–3038. and meta-analysis. Am J Clin Nutr. 2014;100:278–288.
4. Harris W, Mozaffarian D, Lefevre M, et al. Towards establishing dietary reference 27. Sluijs I, Forouhi N, Beulens J, et al. The amount and type of dairy product intake
intakes for eicosapentaenoic and docosahexaenoic acids. J Nutr. 2009;139:804–819. and incident type 2 diabetes: results from the EPIC-InterAct Study. Am J Clin Nutr.
5. Mozaffarian D, Hao T, Rimm EB, et al. Changes in diet and lifestyle and long-term 2012;96:382–390.
weight gain in women and men. N Engl J Med. 2011;364:2392–2404. 28. Mozaffarian D. Effects of dietary fats versus carbohydrates on coronary heart dis-
6. Pan A, Malik VS, Hao T, et al. Changes in water and beverage intake and long- ease: a review of the evidence. Curr Atherosclerosis Rep. 2005;7:435–445.
term weight changes: results from three prospective cohort studies. Int J Obes 29. Mozaffarian R, Lee R, Kennedy M, et al. Identifying whole grain foods: a compari-
(Lond). 2013;37:1378–1385. son of different approaches for selecting more healthful whole grain products.
7. Smith JD, Hou T, Ludwig DS, et al. Changes in intake of protein foods, carbohy- Public Health Nutr. 2013;16:2255–2264.
drate amount and quality, and long-term weight change: results from 3 prospec- 30. Afshin A, Micha R, Khatibzadeh S, et al. Dietary policies to reduce non-
tive cohorts. Am J Clin Nutr. 2015;101:1216–1224. communicable diseases. In: GW Brown, G Yamey, S Wamala, eds. The
8. Bertoia ML, Mukamal KJ, Cahill LE, et al. Changes in Intake of Fruits and Handbook of Global Health Policy. Chichester, UK: John Wiley & Sons, Ltd;
Vegetables and Weight Change in United States Men and Women Followed for 2014:175–193.
Up to 24 Years: Analysis from Three Prospective Cohort Studies. PLoS Med. 31. Centers for Disease Control and Prevention. Achievements in public health, 1900-
2015;12:e1001878. 1999 motor-vehicle safety: A 20th century public health achievement. MMWR
9. Rodrıguez-Ramırez S, Mundo-Rosas V, Garcıa-Guerra A, et al. Dietary patterns are Morb Mortal Wkly Rep. 1999;48:369–374
associated with overweight and obesity in Mexican school-age children. Arch 32. Mozaffarian D, Hemenway D, Ludwig D. Curbing gun violence: lessons from pub-
Latinoam Nutr. 2011;61:270–278. lic health successes. JAMA. 2013;309:551–552.
10. Romero-Polvo A, Denova-Gutiérrez E, Rivera-Paredez B, et al. Association between 33. Mozaffarian D, Afshin A, Benowitz N, et al. Population approaches to improve
dietary patterns and insulin resistance in Mexican children and adolescents. Ann diet, physical activity, and smoking habits a scientific statement from the
Nutr Metabol. 2012;61:142–150. American Heart Association. Circulation. 2012;126:1514–1563.

Nutrition ReviewsV Vol. 75(S1):19–31

R
31